Pyrimidyn Compounds: Dual-Action Small Molecule Pyrimidine-Based ...

Articles pubs.acs.org/acschemicalbiology

Pyrimidyn Compounds: Dual-Action Small Molecule PyrimidineBased Dynamin Inhibitors Andrew B. McGeachie,†,○,# Luke R. Odell,‡,○,# Annie Quan,† James A. Daniel,† Ngoc Chau,† Timothy A. Hill,‡,○ Nick N. Gorgani,† Damien J. Keating,∥ Michael A. Cousin,∥ Ellen M. van Dam,§ Anna Mariana,‡ Ainslie Whiting, Swetha Perera,† Aimee Novelle,† Kelly A. Young,‡ Fiona M. Deane,‡ Jayne Gilbert,¶ Jennette A. Sakoff,¶ Megan Chircop,† Adam McCluskey,‡ and Phillip J. Robinson*,† †

Cell Signalling Unit, Children’s Medical Research Institute, The University of Sydney, Sydney, NSW 2145, Australia Centre for Chemical Biology, Chemistry, The University of Newcastle, Callaghan, NSW 2308, Australia § The Garvan Institute, 384 Victoria Street, Darlinghurst, Sydney, NSW 2010, Australia ∥ Department of Human Physiology, Flinders University, Adelaide, South Australia, 5001, Australia ⊥ Centre for Integrative Physiology, George Square, University of Edinburgh, Edinburgh EH8 9XD, U.K. ¶ Department of Medical Oncology, Calvary Mater Newcastle Hospital, Waratah, NSW 2298, Australia ‡

S Supporting Information *

ABSTRACT: Dynamin is required for clathrin-mediated endocytosis (CME). Its GTPase activity is stimulated by phospholipid binding to its PH domain, which induces helical oligomerization. We have designed a series of novel pyrimidine-based “Pyrimidyn” compounds that inhibit the lipid-stimulated GTPase activity of full length dynamin I and II with similar potency. The most potent analogue, Pyrimidyn 7, has an IC50 of 1.1 μM for dynamin I and 1.8 μM for dynamin II, making it among the most potent dynamin inhibitors identified to date. We investigated the mechanism of action of the Pyrimidyn compounds in detail by examining the kinetics of Pyrimidyn 7 inhibition of dynamin. The compound competitively inhibits both GTP and phospholipid interactions with dynamin I. While both mechanisms of action have been previously observed separately, this is the first inhibitor series to incorporate both and thereby to target two distinct domains of dynamin. Pyrimidyn 6 and 7 reversibly inhibit CME of both transferrin and EGF in a number of non-neuronal cell lines as well as inhibiting synaptic vesicle endocytosis (SVE) in nerve terminals. Therefore, Pyrimidyn compounds block endocytosis by directly competing with GTP and lipid binding to dynamin, limiting both the recruitment of dynamin to membranes and its activation. This dual mode of action provides an important new tool for molecular dissection of dynamin’s role in endocytosis.

E

mechanical release of the vesicle; (ii) the Bundle Signaling Element (BSE)/GTPase effector domain (GED) which appears to be responsible for the self-assembly of dynamin7 and acts as a GTPase activator protein to stimulate GTPase activity;8 (iii) the stalk (also known as the middle domain); (iv) the pleckstrin homology (PH) domain, which targets dynamin to the membrane by interacting with lipids and may also act as a GTPase inhibitory module;9 and (v) the proline-rich domain (PRD), which is involved in the binding of dynamin to a number of SH3 containing proteins as well as the site of in vivo phosphorylation.10,11 In CME, dynamin is recruited to the invaginating vesicle, where assembly into a helical structure occurs around the vesicle neck.12 Dynamin requires its PH domain to target sites of endocytosis5,13,14 that have a hydrophobic loop that inserts into the lipid bilayer for

ndocytosis is a process in which the plasma membrane is selectively deformed inward to form intracellular lipidbound vesicular structures.1 It acts as both a retrieval mechanism for the recycling of membrane and embedded components as well as a means for the capture of extracellular material in the form of internalized cargo.2,3 These vesicles come in a variety of forms, ranging from large phagosomes to small clathrin-coated vesicles, including the smallest, synaptic vesicles. There are a variety of mechanisms for endocytosis, with the best characterized being clathrin-mediated endocytosis (CME) of activated, ligand-bound membrane receptors.1 CME requires a clathrin cage to shape the vesicle size and the large GTPase dynamin II for vesicle neck fission.4 A molecular variant of CME is synaptic vesicle endocytosis (SVE), which is confined to presynaptic nerve terminals and is regulated by dynamin I and utilizes a similar CME machinery but mainly with neuron-specific variants of many of the proteins.5,6 Dynamin consists of five functional domains: (i) the G domain, which binds and hydrolyses GTP resulting in the © 2013 American Chemical Society

Received: February 25, 2013 Accepted: May 3, 2013 Published: May 3, 2013 1507

dx.doi.org/10.1021/cb400137p | ACS Chem. Biol. 2013, 8, 1507−1518

ACS Chemical Biology

Articles

Figure 1. Lead compound 1 and subsequent library design, synthesis, and development of 6 and 7 as potent dynamin I GTPase inhibitors.

membrane curvature generation.15 GTP hydrolysis triggers a conformational constriction, and the vesicle is cut from the plasma membrane.16−18 Many classical inhibitors of endocytosis lack a well-defined mechanism of action.19 Nonspecific inhibitors include cationic amphiphilic drugs (e.g., chlorpromazine),20 concanavalin A, phenylarsine oxide,21 dansylcadaverine,22 intracellular potassium depletion,23 intracellular acidification,24 and decreasing medium temperature to 4 °C.25 Their low potency and lack of a specific target have led to the development of small molecules with defined target proteins, in particular, clathrin26 and dynamin.27 We have developed a number of classes of dynamin inhibitors, including long chain ammonium salts (MiTMAB analogues),28 room-temperature ionic liquids (RTILs),29 Dynole analogues,30,31 Iminodyn analogues,32 Pthaladyn analogues,33 Rhodadyn analogues,34 and the Dyngos,35 which are more potent analogues of dynasore.36 Each inhibitor series is chemically distinct, with different targets within dynamin.27 The inhibitors described to date affect the PH domain and interfere with interaction of dynamin with phospholipids (MiTMAB compounds and RTILs),28,37 the G domain by competing with GTP (Pthaladyn analogues), or allosterically inhibit the GTPase activity (Dynole, Dyngo, Iminodyn and Rhodadyn analogues). We now report a new series of novel dynamin inhibitors called the Pyrimidyn series. They are substituted pyrimidine compounds that also incorporate some design aspects of preexisting dynamin inhibitors that target either the PH domain or the GTPase domain. As the Pyrimidyn analogues are nucleotide

analogues,38−40 they competitively target the GTP binding pocket; however, they also share structural similarities with the ammonium salt inhibitors,28 conferring amphiphilic properties that interfere with dynamin binding to phospholipids. Thus, the Pyrimidyn analogues are the first dual-action dynamin inhibitors.

■

RESULTS AND DISCUSSION

Development of the Pyrimidyn Analogues: Pyrimidine-Based Inhibitors of Dynamin. Prior development of libraries of corticotrophin-releasing hormone antagonists within our laboratories52 resulted in the development of a wide range of substituted pyrimidines.53,54 Given that pyrimidines are known to function as nucleotide mimetics,38−40 we reasoned that these compounds may have utility in the development of inhibitors against the GTP binding site of dynamin. We therefore screened our existing pyrimidine libraries for inhibition of dynamin I and identified Pyrimidyn 1 as a 35.3 ± 7.1 μM dynamin I inhibitor (Supplementary Table S1). Although a relatively poor inhibitor compared to those we had previously developed,28,37,55,56 we were sufficiently encouraged to further develop a targeted library that we refer to as the Pyrimidyn series. Commencing with commercially available 2methyl-4,6-dihydroxypyrimidine, we generated the required 4,6-dichloro analogue. Treatment of this material as previously reported54 at RT with 1 equiv of amine gave rise to an easily separable mixture of monosubstituted 2- and 4-amino pyrimidines (Figure 1). Evaluation of this first iteration highlighted three analogues, Pyrimidyn 2, 3 and 5, with 1.21508



Articles

Figure 2. Pyrimidyn 6 and 7 inhibit the GTPase activity of dynamin I by disrupting PH domain interactions with lipids. (A) The PS-stimulated GTPase activity of native sheep brain dynamin I (WT, 20 nM) was determined in the presence of a range of concentrations of Pyrimidyn 7. An IC50 value of 1.1 μM was obtained from the resulting IC50 curve. (B) In vitro phospholipid binding of WT dynamin in the presence of Pyrimidyn 6 or 7. Normal binding of dynamin to phospholipids is indicated by its presence in the pellet (P) fraction compared to the supernatant (S) fraction. The Coomassie-stained polyacrylamide gel shown is representative of 3 independent experiments. (C) Michaelis−Menten and (D) Lineweaver−Burke plots showing the effect of Pyrimidyn 7 on dynamin I GTPase with increasing concentrations of PS liposomes and GTP concentration (300 μM). This data shows that the compound is a PS-competitive inhibitor of dynamin I. (E) Michaelis−Menten constants Vmax, Km, and Ki (±95% CI) were calculated from triplicate samples performed during a single experiment. Error bars in all graphs represent SEM (n = 3 independent experiments).

to 6-fold increases in dynamin I inhibition relative to the lead compound (Supplementary Table S1). Introduction of the dimethylaminoethyl side chain led to the production of two regioisomers with increased potency in comparison to that of lead Pyrimidyn 1, with Pyrimidyn 3 three times as potent as Pyrimidyn 2. Introduction of a dodecyl side chain as in Pyrimidyn 4 had no effect on dynamin inhibition (this was also observed with the corresponding C4 isomer; data not shown). This suggested that introduction of a second amino substituent may further improve potency. We viewed the synthesis of a Pyrimidyn 2/3 and 5 hybrid as a logical strategy in developing more potent inhibitors. Accordingly, Pyrimidyn 6 and Pyrimidyn 7 were synthesized from Pyrimidyn 4 and Pyrimidyn 5, respectively, via treatment with N,N-dimethylethylenediamine. Evaluation of Pyrimidyn 6 and 7 realized a further 4- to 16-fold increase in dynamin inhibition. The full dose−response curve for Pyrimidyn 7 is shown in Figure 2A. We further examined whether these compounds exhibited equipotent inhibition of both dynamin I and II. IC50 values cannot readily be compared for two different enzymes unless they and their relevant cofactors and activators are present at the same concentrations. To do this, the GTPase assays for dynamin I and II were normalized (modified such that they yielded similar quantities of phosphate released using identical

amounts of dynamin protein). This allows direct comparison of the IC50 values from the two assays. After assay normalization, no selectivity was observed for 6 or 7, with both inhibitors potently inhibiting dynamin I and II with nearly the same potencies (Supplementary Table S2). The Pyrimidyn Series Competitively Inhibit Binding of Both GTP and Phospholipid. We next examined the mechanism of Pyrimidyn 7 inhibition of dynamin I. Phosphatidylserine (PS) binds to the PH domain of dynamin and enhances its GTPase activity57,58 by inducing cooperative helix oligomerization.59 The MiTMAB series of compounds are surface-active, expected to interfere with protein−lipid interactions,60,61 and have been shown to compete for PS binding to the dynamin I PH domain.37 Structural similarities between the MiTMAB and Pyrimidyn series, such as the presence of an extended carbon chain, led us to investigate whether the latter also target the dynamin−phospholipid interaction. In a sedimentation assay, we found that when dynamin was mixed with liposomes and centrifuged, dynamin was located almost entirely in the pelleted lipid fraction (Figure 2B, lanes 4 and 5). Addition of 30 μM concentration of either 6 or 7 did not change this distribution; however, at 100 μM both compounds caused the majority of dynamin to remaining in the supernatant fraction (Figure 2B). This suggests that both 1509



Articles

Figure 3. Pyrimidyn 7 is a competitive dynamin I inhibitor with both GTP and PS. (A) Michaelis−Menten and (B) Lineweaver−Burke plots showing the effect of varying concentrations of 7 on the GTPase activity of WT dynamin I (20 nM), under conditions of increasing GTP concentration and a fixed PS concentration (12.7 μM). This data shows that Pyrimidyn 7 acts as a GTP-competitive inhibitor of dynamin I. (C) Michaelis−Menten constants Vmax, Km, and Ki (±95% CI) were calculated from triplicate samples performed during a single experiment. Pyrimidyns inhibit basal GTPase activity. (D) Effects of different amounts of Pyrimidyn 6 (●) and Pyrimidyn 7 (■) on GTPase activity of GTPase-GED (2000 nM) for 60 min is depicted. All results are representative of at least 2 independent experiments. Error bars represent SEM.

tration of the GTPase-GED construct we needed to use in these assays to remain within the assay’s dynamic range. Therefore the Pyrimidyns inhibit both basal and PS-stimulated activity with approximately the same potency, consistent with the compounds’ GTP-competitive mechanism of action. Pyrimidyn Compounds Block CME in Non-neuronal Cells. Treatment of cells with dynamin inhibitors leads to a decrease in CME.27 The effect of Pyrimidyn compounds on CME was examined by monitoring the uptake of transferrin (Tf) and epidermal growth factor (EGF) in COS7 cells. Both Pyrimidyn 6 and 7 (30 μM) produced a dramatic reduction in EGF-A488 and Tf-TXR internalization (Figure 4A). Pyrimidyn 5 (30 μM), a relatively weak inhibitor of dynamin GTPase activity, had no effect on CME. The same results were observed for EGF-A488 uptake in HER14 cells. The effects were then quantified at different drug concentrations using a semiautomated CME assay. The IC50 for inhibition of CME by Pyrimidyn 7 in COS-7 cells was 12.1 ± 2.1 μM and Pyrimidyn 6 was slightly less potent (19.6 ± 3.5 μM) (Figure 4B). The assay was also carried out using U2OS cells and generated similar IC50 values (Supplementary Table S1). For comparison, the parent Pyrimidyn 1 was ∼20 times less potent than 7 (211 ± 37.1 μM), and 5 had no effect on CME. The observations are in accordance with the relative potency of Pyrimidyn compounds on dynamin GTPase activity, suggesting they block CME by inhibition of dynamin. The Pyrimidyn compounds do not influence total cellular EGF receptor expression levels or EGF receptor activation in A431 cells (Supplementary Figure S1), precluding the possibility that these observations were due to an effect on EGF receptor signaling. The inhibitory effects of Pyrimidyn 6 and 7 on CME were reversible, with 40% recovery after 5 min and 80% recovery after 60 min of removal of the compound (Figure 4C

Pyrimidyn 6 and 7 interfere with the interaction of dynamin with lipids. To investigate whether Pyrimidyn analogues were able to compete with PS for dynamin I binding, thereby inhibiting GTPase activity, we tested the effect of systematically varying the concentration of both Pyrimidyn 7 and PS on the kinetics of dynamin I GTPase activity, while using a fixed concentration of GTP. Addition of increasing concentrations of 7 did not change the Vmax and increased the Km, showing, in combination with Lineweaver−Burke double reciprocal plots, that 7 competitively inhibited the activation of dynamin I by PS (Figure 2C−E). Given that the compounds are pyrimidine analogues, it was also possible that the Pyrimidyns could inhibit dynamin I by competing with GTP binding. Therefore, the kinetics of dynamin I GTPase activity was investigated while varying both the concentration of Pyrimidyn 7 and GTP, while maintaining a constant PS concentration. Pyrimidyn 7 acted in competition with GTP for binding to dynamin (Figure 3A−C). Taken together, these data demonstrated that the Pyrimidyn compounds possess a dual mechanism of activity, inhibiting dynamin interaction with lipid and also competing for GTP binding. We investigated whether the Pyrimidyns were capable of inhibiting basal dynamin activity or whether they were only effective on stimulated dynamin activity. To do this, we performed GTPase assays using the dynamin GTPase-GED construct, which both lacks a PH domain and is incapable of assembling into multimeric structures, therefore exhibiting only basal GTPase activity.44 We found that Pyrimidyns 6 (IC50 = 50 μM) and 7 (IC50 = 85 μM) both inhibited GTPase activity of the GTPase-GED construct (Figure 3D). The IC50 values for both Pyrimidyns using 2000 nM GTPase-GED were ∼40-fold higher than experiments using 20 nM full-length dynamin (Supplementary Table S1), consistent with the high concen1510



Articles

Figure 5. Pyrimidyn compounds prevent the localization of GFP-dyn I-PH domain to the plasma membrane. (A) HeLa cells were transfected with GFP-dyn I-PH domain, which localized to the cytosol and plasma membrane (see lower panel A’ for detail). Pretreatment with Pyrimidyn 6 (B) or Pyrimidyn 7 (C) prevented the construct from localizing to the plasma membrane (shown in detail in the lower panels). This effect was not observed when cells were treated with Pyrimidyn 1 (D). Boxed regions indicate areas of the plasma membrane that are shown at higher magnification in the lower panels (A’−D’). Images are representative of two independent experiments. Scale bar =10 μM.

PLCδ-PH localization to the plasma membrane (Supplementary Figure S2) as previously reported for MiTMAB.37 Pyrimidyn Compounds Inhibit Synaptic Vesicle Turnover at Presynaptic Terminals. Neurotransmitter release during continuous or repeated synaptic depolarization is maintained by exocytosis and compensatory SVE at the presynaptic terminal. To examine the effects of 7 on SVE, a synaptic vesicle (SV) turnover assay was performed in synaptosomes. The styryl dye FM2-10 was used to label SVs as they undergo SVE, which is evoked by a first depolarization (S1). The amount of fluorescent labeling within the synaptosomes is measured in real time, and then the synaptosomes are challenged with a second depolarization (S2), evoking exocytosis of the dye-loaded SVs and resulting in a loss of fluorescence (Figure 6). When Pyrimidyn 7 (30 μM) was present during the dye loading step, a large inhibition of the amount of dye released by the second was observed (24.2 ± 10.9% of control, Figure 6A), which may indicate an inhibition of SVE. However, it is possible that 7 caused a defect in exocytosis and that this exocytic block could explain the lack of dye release from 7-treated synaptic terminals. Therefore, its effect on exocytosis was investigated by measuring stimulated glutamate release from synaptosomes. Synaptosomes were incubated with Pyrimidyn 7 (30 μM) for 10 min and then depolarized using elevated KCl. The compound inhibited Ca2+dependent glutamate release by approximately 40% (58.6 ± 6.5% of control, Figure 6B). The mechanism behind this reduction in exocytosis observed after treatment with 7 is unclear. To resolve whether the decrease in dye loading shown in Figure 6A could be explained solely by a smaller decrease in exocytosis, the retrieval efficiency was calculated.11,48 FM2-10 dye loading was normalized to the total amount of glutamate release for each experimental condition. Under control conditions, the retrieval efficiency is 1, indicating that the amount of exocytosis is balanced by SVE. Values